Thanks to a retrospective analysis of the first 250 cases of sentinel lymph nodes in breast cancer assessed by the authors over the period from October 1998 to December 2000 in the light of a careful review of the literature, it has been possible to establish the importance of careful patient selection, strict compliance with the execution technique and, above all, the need for an adequate learning curve, before the procedure is used in particular protocols and/or in routine clinical practice in the near future. In particular, the training should first of all ensure that all personnel involved, i.e. surgeons, nuclear medicine specialists, and histopathologists, should attend specific courses at qualified Institutions followed by the actual management of a certain number of consecutive cases. In order to perfect the methodology and organisation, a preliminary study in a group of patients at different stages of evolution of the disease is recommended (50 cases in the study population reported). The next phase should include a group of highly selected patients, numbering at least 50-100. The training may be considered complete when in at least 20 cases, an identification rate of at least 90% is achieved with an incidence of false negatives of not more than 5%. In the authors' experience, these results were reached after 100 selected patients, and were later consistently confirmed after a further 50 cases.